Take Home Exercise 3: Prototyping Modules for Geospatial Analytics Shiny Application

1.0 Overview

Demand for Grab’s Ride-hailing services in Jakarta

This project aims to explore the factors affecting demand for Grab’s services in the city of Jakarta through Spatial Interaction Modelling predominantly using Origin-Data Analysis. Our analysis appeals to a diverse set of stakeholders, both for consumers, corporate stakeholders and policy makers especially in Jakarta where traffic congestion is a predominant problem.

Jakarta’s increasing urbanization drives a growing need for on-demand transportation services like Grab, particularly in areas with high concentrations of POIs. However, there is limited understanding of how these points influence traffic congestion across the city. Key questions include:

• Which POIs generate the most ride-hailing traffic?

• How does demand vary across different times of the day, week, and seasons?

• Can ride-hailing services improve access to key locations while alleviating congestion?

Answering these questions will provide insights into Jakarta’s mobility patterns and inform strategies to reduce traffic bottlenecks while maintaining access to key locations.

2.0 Division of Work

My Responsibilities:

1. EDSA

   • Visualization by Time Clusters

   • Visualization by Weather Patterns (Presence or Absence of Rain)

   • Visualization by Most and Least popular spots for Origin and Destination Trips on the District Level

2. OD Analysis (District Level)

   • Flow Maps

   • Origin and Destination Flows by Factor

     • Driving Type

     • Time

     • Weather

   • Push & Pull Analysis

3.0 The data

R Packages Used

pacman::p_load(sf, sp, tidyverse, tmap, spatstat, spdep, leaflet, ggthemes, performance, data.table, reshape2, ggpubr, DT, stplanr, kableExtra, RColorBrewer, dplyr,tidyr, gridExtra, circlize)

3.1 Datasets used

Grab Posisi Data (Grab Posisi Data)

• Contains GPS pings from Grab vehicles, including timestamps, route data, and vehicle type (motorcycle/car).

• Provides insights into ride-hailing demand, traffic hotspots, and movement patterns around key locations in Jakarta.

Supporting Datasets

• Jakarta Points of Interest (POI) (HumData): Includes office buildings, shopping malls, parks, and other key locations.

• Indonesia Population Density Data (ArcGIS): Adds demographic context for understanding mobility trends.

• Weather API (Weatherbit): Supplements our analysis with weather categories to allow for analysis of Grab Demand according to the presence or absence of rain.

• Jakarta Map (HDX): Visualize the geographical districts of the City-State.

3.2 Aspatial Data

1. Category Mapping for Points of Interest

poi_category <- read_csv("data/aspatial/mapping_poi/category_mapping.csv")

Rows: 205 Columns: 2
── Column specification ────────────────────────────────────────────────────────
Delimiter: ","
chr (2): value, category

ℹ Use `spec()` to retrieve the full column specification for this data.
ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.

2. Population (Aggregate by District)

jakarta_population_district <- read_csv("data/aspatial/population/jakarta_township_population.csv") %>% 
  group_by(district) %>% 
  summarize(
    population_2019 = sum(population_2019, na.rm=TRUE)
  ) %>% 
  ungroup()

Rows: 264 Columns: 5
── Column specification ────────────────────────────────────────────────────────
Delimiter: ","
chr (4): province, city, district, township
dbl (1): population_2019

ℹ Use `spec()` to retrieve the full column specification for this data.
ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.

head(jakarta_population_district)

# A tibble: 6 × 2
  district      population_2019
  <chr>                   <dbl>
1 Cakung                  38710
2 Cempaka Putih            7440
3 Cengkareng              40470
4 Cilandak                16880
5 Cilincing               37471
6 Cipayung                21336

We observe that there are 44 Districts within the Area of Jakarta once aggregated to the district level.

2. Trips Data

• Inclusive of Origin, Destination

• Time Cluster

• Weather Conditions

trips <- readRDS("data/rds/trip_data.rds")

colnames(trips)

 [1] "trj_id"                              "driving_mode"                       
 [3] "origin_time"                         "destination_time"                   
 [5] "total_duration_minutes"              "total_distance_km"                  
 [7] "average_speed_kmh"                   "origin_rawlat"                      
 [9] "origin_rawlng"                       "destination_rawlat"                 
[11] "destination_rawlng"                  "origin_lat"                         
[13] "origin_lng"                          "destination_lat"                    
[15] "destination_lng"                     "origin_province"                    
[17] "origin_city"                         "origin_district"                    
[19] "destination_province"                "destination_city"                   
[21] "destination_district"                "origin_datetime"                    
[23] "destination_datetime"                "origin_day"                         
[25] "origin_hour"                         "destination_day"                    
[27] "destination_hour"                    "origin_time_cluster"                
[29] "destination_time_cluster"            "origin_date"                        
[31] "origin_weather_description"          "origin_weather_description_category"

origin_trips <- st_as_sf(trips, coords = c("origin_lng", "origin_lat"), crs=4326)

destination_trips <- st_as_sf(trips, coords = c("destination_lng", "destination_lat"), crs=4326)

3.3 Geospatial Data

1. Read the Indonesia Administrative boundary shapefile

# Step 1: Read the Indonesia administrative boundary shapefile
indonesia <- st_read(
  dsn = "data/geospatial/indo_map", 
  layer = "idn_admbnda_adm3_bps_20200401"
)

Reading layer `idn_admbnda_adm3_bps_20200401' from data source 
  `/Users/jezelei/jezeleii/IS415-GA/Take-Home_Exercise/TakeHomeEx03/data/geospatial/indo_map' 
  using driver `ESRI Shapefile'
Simple feature collection with 7069 features and 16 fields
Geometry type: MULTIPOLYGON
Dimension:     XY
Bounding box:  xmin: 95.01079 ymin: -11.00762 xmax: 141.0194 ymax: 6.07693
Geodetic CRS:  WGS 84

2. Filter for DKI Jakarta and rename it to “Jakarta”
3. Exclude districts belonging to “Kepulauan Seribu”

jakarta_district <- indonesia %>%
  filter(ADM1_EN == "Dki Jakarta") %>%
  mutate(ADM1_EN = "Jakarta")

jakarta_district <- jakarta_district %>%
  filter(ADM2_EN != "Kepulauan Seribu")  # Exclude Kepulauan Seribu

4. We filter down the required columns and rename accordingly

jakarta_district <- jakarta_district %>%
  dplyr::select(ADM1_EN, ADM2_EN, ADM3_EN, geometry)

jakarta_district <- jakarta_district %>%
  rename(
    province = ADM1_EN,
    city = ADM2_EN,
    district = ADM3_EN,
  )

5. We change the CRS to 5580, Indonesia’s CRS
6. We simplify the geometry with a smaller tolerance to reduce subsequent compute load

jakarta_district <- jakarta_district %>%
  st_transform(crs = 5580)  # Transform to WGS84

jakarta_district <- jakarta_district %>%
  mutate(across(where(is.character), tolower))
jakarta_district_df <- jakarta_district %>%
  st_drop_geometry()

jakarta_district <- jakarta_district %>%
  st_simplify(dTolerance = 10.0)  # Smaller tolerance for longitude/latitude data

7. We plot the interactive map using tmap

# Step 8: Plot the interactive map using tmap
tmap_mode("view")  

tmap mode set to interactive viewing

tm_shape(jakarta_district) +  
  tm_polygons("district", palette = "Blues", 
              border.col = "black", lwd = 0.5) +  # Display district polygons with labels
  tm_basemap("OpenStreetMap") 

Warning: Number of levels of the variable "district" is 44, which is larger
than max.categories (which is 30), so levels are combined. Set
tmap_options(max.categories = 44) in the layer function to show all levels.

tmap_mode('plot')

tmap mode set to plotting

Calculation of Centroid:

jakarta_district_centroid <- jakarta_district %>%
  mutate(
    centroid = st_centroid(geometry),
    centroid_lat = st_coordinates(centroid)[, 2],
    centroid_lng = st_coordinates(centroid)[, 1]
 )

jakarta_district_centroid$district <- str_to_title(jakarta_district_centroid$district)

tmap_mode('view')

tmap mode set to interactive viewing

tm_shape(jakarta_district) + 
  tm_polygons(alpha=0.3) + 
  tm_shape(jakarta_district_centroid) + 
  tm_dots(size = 0.1, alpha= 0.5, popup.vars=c("District" = "district"))

tmap_mode('plot')

tmap mode set to plotting

2. Points of Interest Locations

indonesia_poi <- st_read(dsn="data/geospatial/indo_poi", layer="hotosm_idn_points_of_interest_points_shp") %>% st_transform(crs = 5580) 

Reading layer `hotosm_idn_points_of_interest_points_shp' from data source 
  `/Users/jezelei/jezeleii/IS415-GA/Take-Home_Exercise/TakeHomeEx03/data/geospatial/indo_poi' 
  using driver `ESRI Shapefile'
Simple feature collection with 150315 features and 17 fields
Geometry type: POINT
Dimension:     XY
Bounding box:  xmin: 95.04186 ymin: -11.00919 xmax: 141.0188 ymax: 6.073289
Geodetic CRS:  WGS 84

jakarta_poi <- st_intersection(indonesia_poi, jakarta_district)

Warning: attribute variables are assumed to be spatially constant throughout
all geometries

tmap_mode('plot')

tmap mode set to plotting

tm_shape(jakarta_district) + 
  tm_polygons() + 
tm_shape(jakarta_poi) + 
  tm_dots( alpha = 0.2)

4.0 Processing OD Data

For the purposes of this Take-Home Exercise, we will be experimenting with a subset of data, confining the trips dataset to a:

• certain day of week [Friday]

This filtered data will be stored as od_trips_friday for subsequent analysis. After which, we will zoom in on different levels based on the following variables:

1. Weather [rain or not_rain]
2. Vehicle Type [car or motorcycle]
3. Points of Interest within the District [Categorical Count]
4. Time Clusters

4.1 Geospatial Wrangling - Combining Trips Data with Centroid Mapping

1. Filter trips for Friday

od_trips_friday <- trips %>%
  filter(origin_day == "friday")

2. Convert Origin to SF Object
3. Convert Destination to SF object

origin_sf <- od_trips_friday %>%
  st_as_sf(coords = c("origin_lng", "origin_lat"), crs = 5580)

destination_sf <- od_trips_friday %>%
  st_as_sf(coords = c("destination_lng", "destination_lat"), crs = 5580)

4. Transform jakarta_district_centroid to match CRS
5. Perform spatial join for origin and destination points to match with the centroid
6. Remove geometries to perform a left_join as opposed to an st_join

jakarta_district_centroid <- jakarta_district_centroid %>%
  st_transform(crs = 5580)


origin_with_centroids <- st_join(origin_sf, jakarta_district_centroid, left = FALSE) %>%
  select(trj_id, origin_district, driving_mode, origin_time_cluster, 
         origin_weather_description_category, centroid_lat_origin = centroid_lat, 
         centroid_lng_origin = centroid_lng)


destination_with_centroids <- st_join(destination_sf, jakarta_district_centroid, left = FALSE) %>%
  select(trj_id, destination_district, centroid_lat_destination = centroid_lat, 
         centroid_lng_destination = centroid_lng)

7. Merge origin and destination data on trj_id

origin_with_centroids_df <- origin_with_centroids %>% st_set_geometry(NULL)
destination_with_centroids_df <- destination_with_centroids %>% st_set_geometry(NULL)


od_merged <- left_join(origin_with_centroids_df, destination_with_centroids_df, by = "trj_id")

Warning in left_join(origin_with_centroids_df, destination_with_centroids_df, : Detected an unexpected many-to-many relationship between `x` and `y`.
ℹ Row 1948 of `x` matches multiple rows in `y`.
ℹ Row 2720 of `y` matches multiple rows in `x`.
ℹ If a many-to-many relationship is expected, set `relationship =
  "many-to-many"` to silence this warning.

od_merged <- od_merged %>%
  select(trj_id, origin_district, destination_district, driving_mode, origin_time_cluster, 
         origin_weather_description_category, centroid_lat_origin, centroid_lng_origin, 
         centroid_lat_destination, centroid_lng_destination)

head(od_merged)

# A tibble: 6 × 10
  trj_id origin_district   destination_district driving_mode origin_time_cluster
  <chr>  <chr>             <chr>                <chr>        <chr>              
1 10003  kelapa gading     pasar rebo           car          morning lull       
2 10043  setia budi        pasar minggu         car          morning peak       
3 10061  makasar           duren sawit          car          morning peak       
4 10072  setia budi        mampang prapatan     car          morning peak       
5 10086  kebayoran lama    pasar minggu         motorcycle   morning lull       
6 10089  grogol petamburan tanah abang          car          midnight lull      
# ℹ 5 more variables: origin_weather_description_category <chr>,
#   centroid_lat_origin <dbl>, centroid_lng_origin <dbl>,
#   centroid_lat_destination <dbl>, centroid_lng_destination <dbl>

Next, we process the data by filtering out records with undefined destination districts, removing intra-zonal trips and removing duplicate or NA values:

Filtering Out undefined Origin Destination Districs (this could arise because of unmatched districts where there are trips coming to and fro outside of Jakarta). We convert it to lowercase beforehand first.

district_ids <- str_to_lower(str_trim(unique(jakarta_district_centroid$district)))

# Convert `origin_district` and `destination_district` to lowercase and trim whitespace in `od_trips_friday`
od_trips_friday <- od_trips_friday %>%
  mutate(
    origin_district = str_to_lower(str_trim(origin_district)),
    destination_district = str_to_lower(str_trim(destination_district))
  )

# Re-run the unmatched districts check
unmatched_origins <- od_trips_friday %>%
  filter(!origin_district %in% district_ids) %>%
  distinct(origin_district)

unmatched_destinations <- od_trips_friday %>%
  filter(!destination_district %in% district_ids) %>%
  distinct(destination_district)

# Display unmatched IDs
unmatched_origins

# A tibble: 1 × 1
  origin_district   
  <chr>             
1 outside of jakarta

unmatched_destinations

# A tibble: 1 × 1
  destination_district
  <chr>               
1 outside of jakarta  

In the next step, we check for duplicate records as well:

duplicate <- od_merged %>% 
  group_by_all() %>% 
  filter(n() > 1) %>% 
  ungroup()
duplicate

# A tibble: 0 × 10
# ℹ 10 variables: trj_id <chr>, origin_district <chr>,
#   destination_district <chr>, driving_mode <chr>, origin_time_cluster <chr>,
#   origin_weather_description_category <chr>, centroid_lat_origin <dbl>,
#   centroid_lng_origin <dbl>, centroid_lat_destination <dbl>,
#   centroid_lng_destination <dbl>

Since the result is 0, we can proceed with creation of desire line maps after removing the points where there are invalid destination points (which likely correspond with outside_of_jakarta districts)

od_merged<- od_merged %>% 
  drop_na(destination_district)

head(od_merged)

# A tibble: 6 × 10
  trj_id origin_district   destination_district driving_mode origin_time_cluster
  <chr>  <chr>             <chr>                <chr>        <chr>              
1 10003  kelapa gading     pasar rebo           car          morning lull       
2 10043  setia budi        pasar minggu         car          morning peak       
3 10061  makasar           duren sawit          car          morning peak       
4 10072  setia budi        mampang prapatan     car          morning peak       
5 10086  kebayoran lama    pasar minggu         motorcycle   morning lull       
6 10089  grogol petamburan tanah abang          car          midnight lull      
# ℹ 5 more variables: origin_weather_description_category <chr>,
#   centroid_lat_origin <dbl>, centroid_lng_origin <dbl>,
#   centroid_lat_destination <dbl>, centroid_lng_destination <dbl>

5.0 Subset of Prototype: Visualizing Spatial Interaction

5.1 Removing Intra-zonal flows

We filter out rows where origin_district is the same as the destination_district.

od_data <- od_merged %>% 
  filter(origin_district != destination_district)

1. We ensures the district names are lowercased before proceeding to

od_data <- od_data %>%
  mutate(origin_district = tolower(origin_district),
         destination_district = tolower(destination_district)) %>% 
  select(origin_district, destination_district, origin_time_cluster, origin_weather_description_category, everything())

jakarta_district_centroid <- jakarta_district_centroid %>%
  mutate(district = tolower(district))

2. We create origin and destination points with explicit IDs

origin_points <- od_data %>%
  select(origin_district, centroid_lat_origin, centroid_lng_origin) %>%
  distinct() %>%
  st_as_sf(coords = c("centroid_lng_origin", "centroid_lat_origin"), crs = 5580) %>%
  rename(district = origin_district)

destination_points <- od_data %>%
  select(destination_district, centroid_lat_destination, centroid_lng_destination) %>%
  distinct() %>%
  st_as_sf(coords = c("centroid_lng_destination", "centroid_lat_destination"), crs = 5580) %>%
  rename(district = destination_district)

5.2 Creating Desire Lines

desire_lines <- od2line(
  flow = od_data,
  zones = origin_points,
  destination = destination_points
)

colnames(od_data)

 [1] "origin_district"                     "destination_district"               
 [3] "origin_time_cluster"                 "origin_weather_description_category"
 [5] "trj_id"                              "driving_mode"                       
 [7] "centroid_lat_origin"                 "centroid_lng_origin"                
 [9] "centroid_lat_destination"            "centroid_lng_destination"           

od_aggregated <- od_data %>%
  group_by(origin_district, destination_district) %>%
  summarise(trip_count = n_distinct(trj_id), .groups = "drop")

# View the result
head(od_aggregated)

# A tibble: 6 × 3
  origin_district destination_district trip_count
  <chr>           <chr>                     <int>
1 cakung          cempaka putih                 2
2 cakung          cipayung                      2
3 cakung          duren sawit                   5
4 cakung          gambir                        1
5 cakung          jatinegara                    5
6 cakung          johar baru                    1

We save the output in rds format for future use:

write_rds(od_data, "data/rds/od_data.rds")

We import and save the rds into our R environment:

od_data <- read_rds ("data/rds/od_data.rds")

5.2 Visualizing Desire Lines by Time Cluster

We first create a BASEMAP FOR FRIDAY TRIPS

desire_lines_filtered <- desire_lines %>%
  left_join(od_aggregated, by = c("origin_district", "destination_district")) %>% 
  filter(trip_count > 5)

tmap_mode('view')

tmap mode set to interactive viewing

tm_shape(jakarta_district) + 
  tm_polygons(alpha = 0.3) + 
  tm_shape(jakarta_district_centroid) + 
  tm_dots(size = 0.1, 
          alpha = 0.5, 
          popup.vars = c("District" = "district")) +
  tm_shape(desire_lines_filtered) +
  tm_lines(
    col = "trip_count",          
    palette = "viridis",         
    lwd = 1,                    
    alpha = 0.5,                
    popup.vars = c("origin_district", "destination_district", "trip_count")
  ) +
  tm_layout(legend.outside = TRUE)

tmap_mode('plot')

tmap mode set to plotting

For the purposes of dynamic visualization, we avoid sticking with light default colours of tmap and an alternative propsed as seen from above is using the viridis palette.

From here, we can visualize the desire line maps based on the different variables mentioned earlier : 1. Time Cluster 2. Weather Category 3. Vehicle Type

TIME CLUSTER

colnames(od_data)

 [1] "origin_district"                     "destination_district"               
 [3] "origin_time_cluster"                 "origin_weather_description_category"
 [5] "trj_id"                              "driving_mode"                       
 [7] "centroid_lat_origin"                 "centroid_lng_origin"                
 [9] "centroid_lat_destination"            "centroid_lng_destination"           

desire_lines_mid_lull <- desire_lines %>%
  left_join(od_aggregated, by = c("origin_district", "destination_district")) %>% 
  filter(origin_time_cluster == "midnight lull") %>% 
  filter(trip_count > 10)

desire_lines_mid_peak <- desire_lines %>%
  left_join(od_aggregated, by = c("origin_district", "destination_district")) %>% 
  filter(origin_time_cluster == "midnight peak") %>% 
  filter(trip_count > 10)

desire_lines_morn_lull <- desire_lines %>%
  left_join(od_aggregated, by = c("origin_district", "destination_district")) %>% 
  filter(origin_time_cluster == "morning lull") %>% 
  filter(trip_count > 10)

desire_lines_morn_peak <- desire_lines %>%
  left_join(od_aggregated, by = c("origin_district", "destination_district")) %>% 
  filter(origin_time_cluster == "morning peak") %>% 
  filter(trip_count > 10)

desire_lines_aft_lull <- desire_lines %>%
  left_join(od_aggregated, by = c("origin_district", "destination_district")) %>% 
  filter(origin_time_cluster == "afternoon lull") %>% 
  filter(trip_count > 10)

desire_lines_aft_peak <- desire_lines %>%
  left_join(od_aggregated, by = c("origin_district", "destination_district")) %>% 
  filter(origin_time_cluster == "afternoon peak") %>% 
  filter(trip_count > 10)

desire_lines_evn_lull <- desire_lines %>%
  left_join(od_aggregated, by = c("origin_district", "destination_district")) %>% 
  filter(origin_time_cluster == "evening lull") %>% 
  filter(trip_count > 10)

desire_lines_evn_peak <- desire_lines %>%
  left_join(od_aggregated, by = c("origin_district", "destination_district")) %>% 
  filter(origin_time_cluster == "evening peak") %>% 
  filter(trip_count > 10)

tmap_mode('plot')

tmap mode set to plotting

midnight lull

tm_shape(jakarta_district) + 
  tm_polygons(alpha = 0.5) + 
  tm_shape(jakarta_district_centroid) + 
  tm_dots(size = 0.3, 
          alpha = 0.6, 
          popup.vars = c("District" = "district")) +
  tm_shape(desire_lines_mid_lull) +
  tm_lines(
    lwd = "trip_count",       
    scale = c(1,2,5,7),              
    alpha = 0.5,
    popup.vars = c("origin_district", "destination_district", "trip_count")
  ) +
  tm_basemap("OpenStreetMap") + 
  tm_layout(legend.outside = FALSE, 
            main.title="Desire Lines for Grab Trips at Midnight Lull", 
            main.title.position = "center", 
            main.title.size = 0.6, 
            frame=TRUE) + 
  tm_legend(position = c("RIGHT", "BOTTOM"), legend.text.size =0.8)

midnight peak

tm_shape(jakarta_district) + 
  tm_polygons(alpha = 0.5) + 
  tm_shape(jakarta_district_centroid) + 
  tm_dots(size = 0.3, 
          alpha = 0.6, 
          popup.vars = c("District" = "district")) +
  tm_shape(desire_lines_mid_peak) +
  tm_lines(
    lwd = "trip_count",       
    scale = c(1,2,5,7),              
    alpha = 0.5,
    popup.vars = c("origin_district", "destination_district", "trip_count")
  ) +
  tm_basemap("OpenStreetMap") + 
  tm_layout(legend.outside = FALSE, 
            main.title="Desire Lines for Grab Trips at Midnight Peak", 
            main.title.position = "center", 
            main.title.size = 0.6, 
            frame=TRUE) + 
  tm_legend(position = c("RIGHT", "BOTTOM"), legend.text.size =0.8)

morning lull

tm_shape(jakarta_district) + 
  tm_polygons(alpha = 0.5) + 
  tm_shape(jakarta_district_centroid) + 
  tm_dots(size = 0.3, 
          alpha = 0.6, 
          popup.vars = c("District" = "district")) +
  tm_shape(desire_lines_morn_lull) +
  tm_lines(
    lwd = "trip_count",       
    scale = c(1,2,5,7),              
    alpha = 0.5,
    popup.vars = c("origin_district", "destination_district", "trip_count")
  ) +
  tm_basemap("OpenStreetMap") + 
  tm_layout(legend.outside = FALSE, 
            main.title="Desire Lines for Grab Trips at Midnight Lull", 
            main.title.position = "center", 
            main.title.size = 0.6, 
            frame=TRUE) + 
  tm_legend(position = c("RIGHT", "BOTTOM"), legend.text.size =0.8)

morning peak

tm_shape(jakarta_district) + 
  tm_polygons(alpha = 0.5) + 
  tm_shape(jakarta_district_centroid) + 
  tm_dots(size = 0.3, 
          alpha = 0.6, 
          popup.vars = c("District" = "district")) +
  tm_shape(desire_lines_morn_peak) +
  tm_lines(
    lwd = "trip_count",       
    scale = c(1,2,5,7),              
    alpha = 0.5,
    popup.vars = c("origin_district", "destination_district", "trip_count")
  ) +
  tm_basemap("Esri.WorldGrayCanvas") + 
  tm_layout(legend.outside = FALSE, 
            main.title="Desire Lines for Grab Trips at Morning Peak", 
            main.title.position = "center", 
            main.title.size = 0.6, 
            frame=TRUE) + 
  tm_legend(position = c("RIGHT", "BOTTOM"), legend.text.size =0.8)

afternoon lull

tm_shape(jakarta_district) + 
  tm_polygons(alpha = 0.5) + 
  tm_shape(jakarta_district_centroid) + 
  tm_dots(size = 0.3, 
          alpha = 0.6, 
          popup.vars = c("District" = "district")) +
  tm_shape(desire_lines_aft_lull) +
  tm_lines(
    lwd = "trip_count",       
    scale = c(1,2,5,7),              
    alpha = 0.5,
    popup.vars = c("origin_district", "destination_district", "trip_count")
  ) +
  tm_basemap("OpenStreetMap") + 
  tm_layout(legend.outside = FALSE, 
            main.title="Desire Lines for Grab Trips at Afternoon Lull", 
            main.title.position = "center", 
            main.title.size = 0.6, 
            frame=TRUE) + 
  tm_legend(position = c("RIGHT", "BOTTOM"), legend.text.size =0.8)

afternoon peak

tm_shape(jakarta_district) + 
  tm_polygons(alpha = 0.5) + 
  tm_shape(jakarta_district_centroid) + 
  tm_dots(size = 0.3, 
          alpha = 0.6, 
          popup.vars = c("District" = "district")) +
  tm_shape(desire_lines_aft_lull) +
  tm_lines(
    lwd = "trip_count",       
    scale = c(1,2,5,7),              
    alpha = 0.5,
    popup.vars = c("origin_district", "destination_district", "trip_count")
  ) +
  tm_basemap("OpenStreetMap") + 
  tm_layout(legend.outside = FALSE, 
            main.title="Desire Lines for Grab Trips at Afternoon Peak", 
            main.title.position = "center", 
            main.title.size = 0.6, 
            frame=TRUE) + 
  tm_legend(position = c("RIGHT", "BOTTOM"), legend.text.size =0.8)

evening lull

tm_shape(jakarta_district) + 
  tm_polygons(alpha = 0.5) + 
  tm_shape(jakarta_district_centroid) + 
  tm_dots(size = 0.3, 
          alpha = 0.6, 
          popup.vars = c("District" = "district")) +
  tm_shape(desire_lines_evn_lull) +
  tm_lines(
    lwd = "trip_count",       
    scale = c(1,2,5,7,8,9,10,11,12),              
    alpha = 0.5,
    popup.vars = c("origin_district", "destination_district", "trip_count")
  ) +
  tm_basemap("OpenStreetMap") + 
  tm_layout(legend.outside = FALSE, 
            main.title="Desire Lines for Grab Trips at Evening Lull", 
            main.title.position = "center", 
            main.title.size = 0.6, 
            frame=TRUE) + 
  tm_legend(position = c("RIGHT", "BOTTOM"), legend.text.size =0.8)

Legend labels were too wide. Therefore, legend.text.size has been set to 0.71. Increase legend.width (argument of tm_layout) to make the legend wider and therefore the labels larger.

evening peak

tm_shape(jakarta_district) + 
  tm_polygons(alpha = 0.5) + 
  tm_shape(jakarta_district_centroid) + 
  tm_dots(size = 0.3, 
          alpha = 0.6, 
          popup.vars = c("District" = "district")) +
  tm_shape(desire_lines_evn_peak) +
  tm_lines(
    lwd = "trip_count",       
    scale = c(0.5,1,3,7),              
    alpha = 0.5,
    popup.vars = c("origin_district", "destination_district", "trip_count")
  ) +
  tm_basemap("OpenStreetMap") + 
  tm_layout(legend.outside = FALSE, 
            main.title="Desire Lines for Grab Trips at Evening Peak", 
            main.title.position = "center", 
            main.title.size = 0.6, 
            frame=TRUE) + 
  tm_legend(position = c("RIGHT", "BOTTOM"), legend.text.size =0.8)

5.3 Visualizing Desire Lines by Weather Category

For urban planners, policy makers and corporate stakeholders like Grab which rely on weather data to determine prices (where we hypothesize that strong rain for example will lead to higher demand of Grab vehicles)

From here, we can visualize the desire line maps based on the different variables mentioned earlier : 1. Time Cluster 2. Weather Category 3. Vehicle Type

WEATHER CATEGORY

desire_lines_not_rain <- desire_lines %>%
  left_join(od_aggregated, by = c("origin_district", "destination_district")) %>% 
  filter(origin_weather_description_category == "not_rain") %>%   filter(trip_count > 10)

desire_lines_rain <- desire_lines %>%
  left_join(od_aggregated, by = c("origin_district", "destination_district")) %>% 
  filter(origin_weather_description_category == "rain") %>% 
  filter(trip_count > 10)

No Rain

tmap_mode('view')

tmap mode set to interactive viewing

tm_shape(jakarta_district) + 
  tm_polygons(alpha = 0.2) + 
  tm_shape(jakarta_district_centroid) + 
  tm_dots(size = 0.1, 
          alpha = 0.6, 
          popup.vars = c("District" = "district")) +
  tm_shape(desire_lines_not_rain) +
  tm_lines(
    lwd = "trip_count",       
    scale = c(0.5,1,3,7),              
    alpha = 0.5,
    popup.vars = c("origin_district", "destination_district", "trip_count")
  ) +
  tm_basemap("CartoDB.Positron") + 
  tm_layout(legend.outside = TRUE, 
            main.title="Desire Lines for No Rain Conditions", 
            main.title.position = "center", 
            main.title.size = 0.6, 
            frame=TRUE) + 
   tm_view(view.legend.position = c("RIGHT", "BOTTOM"))

Legend for line widths not available in view mode.

tmap_mode('plot')

tmap mode set to plotting

Note: In our data processing, we aggregated weather conditions into rain and no rain for the purposes of simplification for the exercise and to optimize data processing.

Rain

tmap_mode('view')

tmap mode set to interactive viewing

tm_shape(jakarta_district) + 
  tm_polygons(alpha = 0.2) + 
  tm_shape(jakarta_district_centroid) + 
  tm_dots(size = 0.1, 
          alpha = 0.6, 
          popup.vars = c("District" = "district")) +
  tm_shape(desire_lines_rain) +
  tm_lines(
    lwd = "trip_count",       
    scale = c(0.5,1,3,7),              
    alpha = 0.5,
    popup.vars = c("origin_district", "destination_district", "trip_count")
  ) +
  tm_basemap("CartoDB.Positron") + 
  tm_layout(legend.outside = TRUE, 
            main.title="Desire Lines for Rain Conditions", 
            main.title.position = "center", 
            main.title.size = 0.6, 
            frame=TRUE) 

Legend for line widths not available in view mode.

tmap_mode('plot')

tmap mode set to plotting

5.3.1 Graphical Distribution of Weather Conditions throughout the Time Clusters:

Besides Map interactivity, a useful visualization for new and expert users alike are accompanying bar charts to show the weather distribution for the selected day of week, in this case Fridays.

The code chunk below 1. We summarize the origin_time_cluster and origin_weather_description category 2. Plot accordingly in a comparative bar graph

rain_distribution <- od_merged %>%
  group_by(origin_time_cluster, origin_weather_description_category) %>%
  summarize(count = n(), .groups = "drop") %>%
  ungroup() %>%
 
  complete(origin_time_cluster, origin_weather_description_category, fill = list(count = 0))

ggplot(rain_distribution, aes(y = origin_time_cluster, x = count, fill = origin_weather_description_category)) +
  geom_bar(stat = "identity", position = "dodge") +
  labs(
    title = "Distribution of Rain and Not Rain Conditions by Time Cluster",
    y = "Time Cluster",
    x = "Count",
    fill = "Weather Condition"
  ) +
  theme_minimal() +
  theme(axis.text.y = element_text(angle = 0, hjust = 1))

5.4 Visualizing Desire Lines by Vehicle Type

Upon further inspection of the data, a key perspective we can investigate is the distribution of trips or how vehicle type affects desire lines and demand for Grab Vehicles.

Given Indonesia’s unique context of Traffic Congestion where to curb this, it has implemented legislation such as the Odd-Even traffic policy(TL;DR: License plates ending in odd-numbers can only drive in odd-numbered days, and the same goes for even numbers). This policy was eventually exempted for Ride-hailing services such as Grab and Gojek, but such policies shed light on the dire state of traffic congestion within the country.

There is a general understanding that motorcycle taxis in traffic-congested areas within Indonesia, especially its capital, Jakarta, are much faster on the road given they can pass through narrow roads and is much more accessible in terms of price for the average Indonesian. Given its context of one of the most traffic-congested countries in the world, the prioritization of convenience and accessibility has made the use of motorbikes more prevalent in Indonesia, as compared to other countries such as Singapore where reliable public transportation may offset such demand.

The code chunk below shows the distribution of trips by vehicle type ‘cars’ and ‘motorcycle’ respectively

desire_lines_car <- desire_lines %>%
  left_join(od_aggregated, by = c("origin_district", "destination_district")) %>% 
  filter(driving_mode == "car") %>% 
  filter(trip_count > 10)

desire_lines_motorcycle <- desire_lines %>%
  left_join(od_aggregated, by = c("origin_district", "destination_district")) %>% 
  filter(driving_mode == "motorcycle") %>% 
  filter(trip_count > 10)


tmap_mode('plot')

tmap mode set to plotting

Car

tm_shape(jakarta_district) + 
  tm_polygons(alpha = 0.2) + 
  tm_shape(jakarta_district_centroid) + 
  tm_dots(size = 0.1, 
          alpha = 0.6, 
          popup.vars = c("District" = "district")) +
  tm_shape(desire_lines_car) +
  tm_lines(
    lwd = "trip_count",       
    scale = c(0.5,1,3,7),              
    alpha = 0.5,
    popup.vars = c("origin_district", "destination_district", "trip_count")
  ) +
  tm_basemap("CartoDB.Positron") + 
  tm_layout(legend.outside = TRUE, 
            main.title="Desire Lines for Driving Mode: Car", 
            main.title.position = "center", 
            main.title.size = 0.6, 
            frame=TRUE) + 
  tm_legend(position = c("RIGHT", "BOTTOM"), legend.text.size =0.8)

Motorcycle

tm_shape(jakarta_district) + 
  tm_polygons(alpha = 0.2) + 
  tm_shape(jakarta_district_centroid) + 
  tm_dots(size = 0.1, 
          alpha = 0.6, 
          popup.vars = c("District" = "district")) +
  tm_shape(desire_lines_motorcycle) +
  tm_lines(
    lwd = "trip_count",       
    scale = c(0.5,1,3,7),              
    alpha = 0.5,
    popup.vars = c("origin_district", "destination_district", "trip_count")
  ) +
  tm_basemap("CartoDB.Positron") + 
  tm_layout(legend.outside = TRUE, 
            main.title="Desire Lines for Driving Mode: Motorcycle", 
            main.title.position = "center", 
            main.title.size = 0.6, 
            frame=TRUE) + 
  tm_legend(position = c("RIGHT", "BOTTOM"), legend.text.size =0.8)

5.5 Visualizing Distribution of POIs

Here, we can aggregate the counts of the POIs according to their categories based on their geographic distribution. We can use the centroid of the district we have calculated earlier to combine trips and jakarta_district data

1. We perform spatial intersection to associate POIs with Jakarta Districts
2. Join poi_category to get the category column
3. Aggregate by district and category

We join the poi_category table with the jakarta_poi dataset to map the columns accordingly:

poi_jakarta_district <- st_intersection(jakarta_poi, jakarta_district) %>% 
  select(district, value = amenity) %>% 
  st_drop_geometry() 

Warning: attribute variables are assumed to be spatially constant throughout
all geometries

poi_category_jakarta <- poi_jakarta_district %>% 
  left_join(poi_category, by = "value")

poi_aggregated <- poi_category_jakarta %>% 
  group_by(district, category) %>% 
  summarise(count = n(), .groups='drop') %>% 
  filter(!is.na(category))

datatable(poi_aggregated)

We can visualize the distribution of the POIs

poi_aggregated_wide <- poi_aggregated %>%
  pivot_wider(
    names_from = category, 
    values_from = count, 
    values_fill = 0  # Fill missing values with 0
  )

jakarta_district_counts <- jakarta_district %>%
  left_join(poi_aggregated_wide, by = "district")

category_maps <- names(poi_aggregated_wide)[-1]  # Exclude 'district' column
map_list <- lapply(category_maps, function(category) {
  tm_shape(jakarta_district_counts) +
    tm_polygons(
      col = category, 
      palette = "Blues", 
      title = paste(category),
      style = "quantile"
    ) +
    tm_layout(main.title = paste(category), main.title.size = 1.2)
})

We establish common breaks to standardize the legend values across the different maps as opposed to taking the automatic legend values based on quantiles per map. This is to facilitate ease of comparison and visualization for the User.

common_breaks <- c(0, 20, 40, 60, 80, 100, max(poi_aggregated_wide[-1], na.rm = TRUE))


map_list <- lapply(category_maps, function(category) {
  tm_shape(jakarta_district_counts) +
    tm_polygons(
      col = category, 
      palette = "Blues", 
      title = paste(category),
      breaks = common_breaks  
    ) +
    tm_layout(main.title = paste(category), main.title.size = 1.2)
})

tmap_arrange(map_list[[1]], map_list[[2]], map_list[[3]], map_list[[4]], ncol = 2, nrow = 2)

tmap_arrange(map_list[[5]], map_list[[6]], map_list[[7]], map_list[[8]], ncol = 2, nrow = 2)

We can also visualize it using the tooltip of each district’s centroid we have calculated earlier. We join the tables for jakarta_district_centroid and the POI mapping and append a tooltip column

colnames(jakarta_district_centroid)

[1] "province"     "city"         "district"     "geometry"     "centroid"    
[6] "centroid_lat" "centroid_lng"

unique(poi_category$category)

[1] "Facilities_Services"      "Essentials"              
[3] "Offices_Business"         "Cultural_Attractions"    
[5] "Restaurants_Food"         "Recreation_Entertainment"
[7] "Others"                   "Shops"                   
[9] "Tourism_Spots"           

1. Ensure district columns are in lowercase in both dataframes

poi_aggregated <- poi_aggregated %>%
  mutate(district = tolower(district))

jakarta_district_centroid <- jakarta_district_centroid %>%
  mutate(district = tolower(district))

2. Pivot poi_aggregated to create a separate column for each category

tooltip_data <- poi_aggregated %>%
  pivot_wider(
    names_from = category,
    values_from = count,
    values_fill = list(count = 0)  # Fill missing values with 0
  )

3. Ensure the columns are within tooltip_data, for values with NA (categories that are not present for that district), we will with 0

required_categories <- c(
  "Facilities_Services", "Restaurants_Food", "Essentials", 
  "Offices_Business", "Cultural_Attractions", "Recreation_Entertainment",
  "Shops", "Tourism_Spots", "Others"
)

for (category in required_categories) {
  if (!category %in% colnames(tooltip_data)) {
    tooltip_data[[category]] <- 0
  }
}

jakarta_district_centroid_expanded <- jakarta_district_centroid %>%
  left_join(tooltip_data, by = "district")

4. We plot the map

tmap_mode("view")

tmap mode set to interactive viewing

tm_shape(jakarta_district) +  
  tm_polygons(alpha = 0.3, border.col = "black") +  # District boundaries
  tm_shape(jakarta_district_centroid_expanded) + 
  tm_dots(
    size = 0.1, col = "blue", alpha = 0.5,
    popup.vars = c(
      "District" = "district",
      "Facilities_Services" = "Facilities_Services",
      "Restaurants_Food" = "Restaurants_Food",
      "Essentials" = "Essentials",
      "Offices_Business" = "Offices_Business",
      "Cultural_Attractions" = "Cultural_Attractions",
      "Recreation_Entertainment" = "Recreation_Entertainment", 
      "Shops" = "Shops", 
      "Tourism_Spots" = "Tourism_Spots",
      "Others" = "Others"
    )
  ) +
  tm_basemap("OpenStreetMap")

tmap_mode("plot")

tmap mode set to plotting

Now that we can see the number of POI categories per district by clicking into the centroid, we can also visualize this distribution through exploring the top 10 districts with the most POI to potentially use for subsequent analysis:

1. We obtain the top_districts by summing up the count across all categories per district
2. We filter to the top 5 districts
3. Arrange the Category count by descending order per district (using )
4. Create a Grouped Bar Chart

top_districts <- poi_aggregated %>%
  group_by(district) %>%
  summarise(total_count = sum(count)) %>%
  arrange(desc(total_count)) %>%
  slice(1:5) %>%
  pull(district)


poi_aggregated_top <- poi_aggregated %>%
  filter(district %in% top_districts)


spacing_rows <- poi_aggregated_top %>%
  distinct(district) %>%
  mutate(category = NA, count = NA)


poi_aggregated_with_spacing <- bind_rows(poi_aggregated_top, spacing_rows) %>%
  arrange(district)

#create a grouped bar chart 
ggplot(poi_aggregated_with_spacing, aes(y = district, x = count, fill = category)) +
  geom_bar(stat = "identity", position = "dodge", na.rm = TRUE) +
  labs(
    title = "Distribution of POI Categories by District",
    x = "Count",
    y = "District",
    fill = "Category"
  ) +
  theme_minimal() +
  theme(
    axis.text.y = element_text(angle = 0, hjust = 1),
    legend.position = "bottom",
    plot.margin = margin(10, 10, 10, 10)
  ) +
  scale_fill_brewer(palette = "Set3")

5.6 Prototype Samples for Shiny App

After exploring the various different variables we can filter with, this would be our sample interface by filtering with the above variables for OD Analysis - Day of Week - Time Cluster - Weather Conditions - Vehicle Type.

With this as our base template:

6.0 Spatial Interaction Modelling

Aside from the graphical representation of the desire lines with respect to the selected variables, we will also explore the implementation of interaction models of OD data of the Grab Trips.

We will be exploring SIM to determine factors affecting Grab Vehicle demand flows during Fridays in the our Grab Dataset which spans 2 weeks in Jakarta.

We will be exploring

1. Origin Constrained Model
2. Destination Constrained Model
3. Doubly Constrained Model

head(poi_aggregated)

# A tibble: 6 × 3
  district category             count
  <chr>    <chr>                <int>
1 cakung   Cultural_Attractions     2
2 cakung   Essentials              12
3 cakung   Facilities_Services     22
4 cakung   Offices_Business        18
5 cakung   Others                   1
6 cakung   Restaurants_Food         3

6.1 Computing Distance Matrix

The computed distance matrix shows the distance between pairs of locations, in this case the distance between origin district and destination districts of each trajectory of the trips. The diagonal of the table will demonstrate that a location’s distance from itself (e.g origin district distance from origin distrct) will be 0.

We will use the jakarta_district_centroid_expanded to compute the distance of each district from each other

jakarta_dist_centroid_sp <- as(jakarta_district_centroid_expanded$centroid, "Spatial")
jakarta_dist_centroid_sp

class       : SpatialPoints 
features    : 44 
extent      : 13484362, 13585052, -3018643, -2900311  (xmin, xmax, ymin, ymax)
crs         : +proj=tmerc +lat_0=0 +lon_0=30 +k=1 +x_0=300000 +y_0=0 +ellps=krass +units=m +no_defs 

For the purposes of optimizing data processing for our final Shiny application, we will process using sp method as compared to sf, due to less compute power generally needed for sp functions.

We use spDists() of sp package to compute the distance between centroids of the Jakarta Districts

dist_districts <- spDists(jakarta_dist_centroid_sp, longlat=FALSE)
head(dist_districts, n=c(5,5))

         [,1]     [,2]     [,3]      [,4]      [,5]
[1,]     0.00 33262.49 94446.12  81967.21  28291.84
[2,] 33262.49     0.00 61294.34  61065.20  42488.85
[3,] 94446.12 61294.34     0.00  68589.05  95302.01
[4,] 81967.21 61065.20 68589.05      0.00 101922.70
[5,] 28291.84 42488.85 95302.01 101922.70      0.00

For ease of visualization, we can use the kableExtra package to plot a highlight table for the distance matrix

Before which, we convert it into a data frame and label the rows and columns with district names

dist_districts_df <- as.data.frame(dist_districts)

rownames(dist_districts) <- jakarta_district_centroid_expanded$district
colnames(dist_districts) <- jakarta_district_centroid_expanded$district

At this juncture, we represent the distance matrix calculated to display using the Highlight table of distances:

In our final prototype, we also consider cognitive load and visual comprehension so we will mock some changes we might implement, such as rounding to the nearest integer, using a sequential colour scale (where higher distance values represent longer distances as well as ease of reading the data)

1. Round the Distance to the nearest integer

rounded_distances <- dist_districts_df[1:44, 1:44] %>%
  mutate(across(everything(), round))

2. Define a sequential color palette (using RBrewer)

color_palette <- colorRampPalette(brewer.pal(9, "Blues"))(100)

3. Apply the color palette to the distance values. As values above 70k can become dark, we will conditionally make it into a white font

highlighted_distances <- rounded_distances %>%
  mutate(across(everything(),
                ~ cell_spec(., 
                            color = ifelse(. > 90000, "white", "black"),
                            background = spec_color(as.numeric(.),
                                                    option = "custom",
                                                    scale_from = range(rounded_distances, na.rm = TRUE),
                                                    palette = color_palette),
                            bold = TRUE)))

4. We create the Highlight table

dist_matrix_highlight <- kable(highlighted_distances, escape = FALSE, format = "html") %>%
  kable_styling("striped", full_width = TRUE) %>%
  add_header_above(c(" " = 1, "Distance Matrix" = 43))
dist_matrix_highlight

	Distance Matrix
V1	V2	V3	V4	V5	V6	V7	V8	V9	V10	V11	V12	V13	V14	V15	V16	V17	V18	V19	V20	V21	V22	V23	V24	V25	V26	V27	V28	V29	V30	V31	V32	V33	V34	V35	V36	V37	V38	V39	V40	V41	V42	V43	V44
0	33262	94446	81967	28292	64522	71799	41186	46710	21978	56386	71165	85196	33815	39344	110332	68270	78067	78093	21166	40388	90523	34965	52479	41945	61664	35340	48144	54441	66063	53535	68673	73567	85634	87349	23632	51485	43230	52353	58752	65331	60279	39679	44837
33262	0	61294	61065	42489	67665	69558	14199	17160	33560	23163	37989	72555	21660	6187	77380	41951	50687	45205	18635	9920	57913	32082	43320	40315	38883	10482	16443	27423	33357	35603	52897	66571	54358	58887	11079	20098	10233	24744	27118	33178	28989	22603	23202
94446	61294	0	68589	95302	109158	104370	55554	49643	91374	38144	23305	90985	72626	55419	16643	51240	48269	22466	76241	54588	21028	78107	81172	87355	59384	62745	49051	47031	32294	67487	75538	96151	20645	45601	71876	44048	51852	52305	37269	30531	40980	63767	62486
81967	61065	68589	0	101923	52073	42693	70717	68651	63941	56096	58637	22807	48268	57361	82192	22018	20443	47983	78175	65452	48598	93131	33577	45999	22185	51750	47868	76817	45737	28445	16326	32663	80022	24670	62653	65557	54665	37047	66467	66034	40162	81983	38946
28292	42489	95302	101923	0	92357	98806	40571	46237	49397	60903	73727	109298	55043	47940	109216	84218	93154	84638	24028	43390	97610	18025	76833	68013	80001	50378	58713	48779	73805	74230	91027	99382	81082	101343	39341	52004	51834	66998	58186	64820	71057	31603	62977
64522	67665	109158	52073	92357	0	12031	81827	83690	42970	79964	90718	37017	46030	68824	125149	58997	65677	86716	74980	77036	93191	93129	28247	27383	50022	58114	66548	94176	77968	41821	37138	21178	112833	73858	60456	84279	69277	58512	89326	92784	69203	89093	49975
71799	69558	104370	42693	98806	12031	0	83384	84422	49827	78503	87715	25368	48175	69650	119848	53240	58564	81960	79501	78274	87007	97553	26720	31087	45104	59368	65541	94737	74472	38829	29504	10174	110152	65923	63969	84187	69397	56186	88460	91116	65987	91820	49166
41186	14199	55554	70717	40571	81827	83384	0	5919	46632	20574	33261	84596	35799	14182	70515	49784	57102	44243	21053	5402	57180	25307	56888	54500	48996	24017	22888	13961	34044	47796	64729	79863	44532	63929	23685	11568	16102	33713	18379	25167	33315	11624	35487
46710	17160	49643	68651	46237	83690	84422	5919	0	50596	15650	27495	83848	37859	14877	64610	47172	53795	38978	26905	7270	51840	30335	57729	56598	47472	25591	21035	10501	29390	47540	63989	80226	38924	60048	27729	5768	15268	32115	12465	19251	29854	15967	35608
21978	33560	91374	63941	49397	42970	49827	46632	50596	0	54318	68662	64179	19806	37840	107956	54456	64308	71632	33656	43330	82693	51405	31960	20403	46187	28729	42323	60513	59825	36481	49193	51807	87367	73985	22960	53565	40508	41691	60371	65966	52100	50451	31842
56386	23163	38144	56096	60903	79964	78503	20574	15650	54318	0	14843	73712	37120	17294	54345	34097	39364	23737	39294	17675	36707	45810	52182	54482	36699	25697	13517	21089	13766	39801	54324	72720	33741	44844	33779	10580	13920	22511	10544	12833	15961	31595	29995
71165	37989	23305	58637	73727	90718	87715	33261	27495	68662	14843	0	78958	50583	32131	39545	37484	39099	14216	53399	31625	25700	57379	62485	66752	43230	39933	26589	27838	13493	49125	60876	80734	22482	41409	48607	21731	28640	32240	16026	10603	21729	42814	41657
85196	72555	90985	22807	109298	37017	25368	84596	83848	64179	73712	78958	0	54400	70459	104925	41491	42729	69705	87002	79194	71318	103904	32762	43823	37019	62190	62941	93190	65562	36973	19879	15968	101033	47438	70716	81897	68765	51962	84253	85039	58510	94947	49404
33815	21660	72626	48268	55043	46030	48175	35799	37859	19806	37120	50583	54400	0	23133	89268	35379	45290	52169	32717	31045	62952	50185	22725	18758	27994	12357	24019	48359	40594	19838	36035	46111	70847	54870	16465	39107	24356	21898	44994	49629	32529	43651	12138
39344	6187	55419	57361	47940	68824	69650	14182	14877	37840	17294	32131	70459	23133	0	71633	37306	45600	39018	24408	8815	51734	36134	43010	41808	35339	10785	10790	25353	27183	33617	50609	65791	49537	53459	16489	16134	4074	20403	22593	28156	23162	24756	21313
110332	77380	16643	82192	109216	125149	119848	70515	64610	107956	54345	39545	104925	89268	71633	0	66612	62399	38550	91482	70120	33607	91560	97347	103877	75175	79286	65642	60443	48902	83680	90657	111280	29146	58038	88122	59245	68185	68860	52149	45375	57533	77642	79082
68270	41951	51240	22018	84218	58997	53240	49784	47172	54456	34097	37484	41491	35379	37306	66612	0	9911	29013	60191	44712	34196	73438	32362	41945	9492	34213	26991	54930	24172	19764	24342	44920	59569	19538	46166	43741	34083	17232	44449	44184	18148	61298	23521
78067	50687	48269	20443	93154	65677	58564	57102	53795	64308	39364	39099	42729	45290	45600	62399	9911	0	27614	69177	52310	28884	81516	40570	51025	18573	43633	34912	60446	27091	28967	29066	49381	59851	9697	55683	49676	42054	26348	49078	47494	23941	68722	33402
78093	45205	22466	47983	84638	86716	81960	44243	38978	71632	23737	14216	69705	52169	39018	38550	29013	27614	0	62644	41329	12976	69313	58809	65555	36927	43863	30182	41348	12084	45135	53354	73904	32387	28017	54574	33377	34998	30810	29163	24669	19664	54886	41163
21166	18635	76241	78175	24028	74980	79501	21053	26905	33656	39294	53399	87002	32717	24408	91482	60191	69177	62644	0	21781	75558	18154	55326	48472	56131	26460	35043	33559	51195	50965	68088	78562	65457	77503	16289	32261	28433	42974	39341	46108	47545	18608	39282
40388	9920	54588	65452	43390	77036	78274	5402	7270	43330	17675	31625	79194	31045	8815	70120	44712	52310	41329	21781	0	54303	29402	51702	49800	43653	18989	17731	17523	30448	42410	59327	74580	45522	59458	20493	11123	10788	28415	18393	24945	28751	16595	30127
90523	57913	21028	48598	97610	93191	87007	57180	51840	82693	36707	25700	71318	62952	51734	33607	34196	28884	12976	75558	54303	0	82166	66222	74488	43377	55769	42394	53505	24560	52897	57575	78105	38044	24781	66914	46182	47681	41087	41489	36269	30776	67687	51307
34965	32082	78107	93131	18025	93129	97553	25307	30335	51405	45810	57379	103904	50185	36134	91560	73438	81516	69313	18154	29402	82166	0	72910	66576	70955	41951	46628	31182	59350	67228	84542	96272	63183	88853	33789	35990	39478	56464	41468	47875	58145	14566	55007
52479	43320	81172	33577	76833	28247	26720	56888	57729	31960	52182	62485	32762	22725	43010	97347	32362	40570	58809	55326	51702	66222	72910	0	12504	22923	32948	38992	68022	49737	13685	17699	23413	84669	49781	39156	57486	42677	30325	61928	64930	40960	65836	22493
41945	40315	87355	45999	68013	27383	31087	54500	56598	20403	54482	66752	43823	18758	41808	103877	41945	51025	65555	48472	49800	74488	66576	12504	0	32556	31028	40980	67100	55127	22182	30201	31642	88074	60555	33289	57608	42669	35062	63146	67260	46377	61767	25109
61664	38883	59384	22185	80001	50022	45104	48996	47472	46187	36699	43230	37019	27994	35339	75175	9492	18573	36927	56131	43653	43377	70955	22923	32556	0	29671	26440	56396	29768	10409	17809	37555	65694	28225	40938	45065	32895	15368	47237	48275	21861	60064	17206
35340	10482	62745	51750	50378	58114	59368	24017	25591	28729	25697	39933	62190	12357	10785	79286	34213	43633	43863	26460	18989	55769	41951	32948	31028	29671	0	13815	36089	31814	25316	42640	56124	59170	52544	12095	26771	12232	17347	32828	37801	25122	33084	13068
48144	16443	49051	47868	58713	66548	65541	22888	21035	42323	13517	26589	62941	24019	10790	65642	26991	34912	30182	35043	17731	42394	46628	38992	40980	26440	13815	0	30298	18100	27292	43134	60334	47134	42670	24522	19110	7150	11080	22937	26337	12562	34326	16592
54441	27423	47031	76817	48779	94176	94737	13961	10501	60513	21089	27838	93190	48359	25353	60443	54930	60446	41348	33559	17523	53505	31182	68022	67100	56396	36089	30298	0	34142	57387	73420	90278	32595	65580	37553	11336	25364	41229	12221	17469	36918	17278	45729
66063	33357	32294	45737	73805	77968	74472	34044	29390	59825	13766	13493	65562	40594	27183	48902	24172	27091	12084	51195	30448	24560	59350	49737	55127	29768	31814	18100	34142	0	36174	47383	67306	35927	31443	42511	24296	23124	20069	22275	20451	8826	45288	30192
53535	35603	67487	28445	74230	41821	38829	47796	47540	36481	39801	49125	36973	19838	33617	83680	19764	28967	45135	50965	42410	52897	67228	13685	22182	10409	25316	27292	57387	36174	0	17324	33050	71463	38592	34921	46354	32283	17368	49970	52317	27460	57980	12431
68673	52897	75538	16326	91027	37138	29504	64729	63989	49193	54324	60876	19879	36035	50609	90657	24342	29066	53354	68088	59327	57575	84542	17699	30201	17809	42640	43134	73420	47383	17324	0	20633	83281	36772	51903	62161	48887	32234	64826	66071	39665	75171	29697
73567	66571	96151	32663	99382	21178	10174	79863	80226	51807	72720	80734	15968	46111	65791	111280	44920	49381	73904	78562	74580	78105	96272	23413	31642	37555	56124	60334	90278	67306	33050	20633	0	103214	56298	62489	79353	64981	50209	82987	85029	59142	89147	44619
85634	54358	20645	80022	81082	112833	110152	44532	38924	87367	33741	22482	101033	70847	49537	29146	59569	59851	32387	65457	45522	38044	63183	84669	88074	65694	59170	47134	32595	35927	71463	83281	103214	0	60079	65436	34450	47002	54344	27262	21411	44176	49831	63213
87349	58887	45601	24670	101343	73858	65923	63929	60048	73985	44844	41409	47438	54870	53459	58038	19538	9697	28017	77503	59458	24781	88853	49781	60555	28225	52544	42670	65580	31443	38592	36772	56298	60079	0	64637	55407	49694	35200	53676	50962	30709	75516	42894
23632	11079	71876	62653	39341	60456	63969	23685	27729	22960	33779	48607	70716	16465	16489	88122	46166	55683	54574	16289	20493	66914	33789	39156	33289	40938	12095	24522	37553	42511	34921	51903	62489	65436	64637	0	31124	20063	29438	38190	44241	36677	28779	23765
51485	20098	44048	65557	52004	84279	84187	11568	5768	53565	10580	21731	81897	39107	16134	59245	43741	49676	33377	32261	11123	46182	35990	57486	57608	45065	26771	19110	11336	24296	46354	62161	79353	34450	55407	31124	0	15003	29939	7272	13908	25883	21512	35026
43230	10233	51852	54665	51834	69277	69397	16102	15268	40508	13920	28640	68765	24356	4074	68185	34083	42054	34998	28433	10788	47681	39478	42677	42669	32895	12232	7150	25364	23124	32283	48887	64981	47002	49694	20063	15003	0	17628	20643	25604	19223	27343	20380
52353	24744	52305	37047	66998	58512	56186	33713	32115	41691	22511	32240	51962	21898	20403	68860	17232	26348	30810	42974	28415	41087	56464	30325	35062	15368	17347	11080	41229	20069	17368	32234	50209	54344	35200	29438	29939	17628	0	32839	34949	11328	44937	10359
58752	27118	37269	66467	58186	89326	88460	18379	12465	60371	10544	16026	84253	44994	22593	52149	44449	49078	29163	39341	18393	41489	41468	61928	63146	47237	32828	22937	12221	22275	49970	64826	82987	27262	53676	38190	7272	20643	32839	0	6788	26312	26919	39486
65331	33178	30531	66034	64820	92784	91116	25167	19251	65966	12833	10603	85039	49629	28156	45375	44184	47494	24669	46108	24945	36269	47875	64930	67260	48275	37801	26337	17469	20451	52317	66071	85029	21411	50962	44241	13908	25604	34949	6788	0	26530	33392	42822
60279	28989	40980	40162	71057	69203	65987	33315	29854	52100	15961	21729	58510	32529	23162	57533	18148	23941	19664	47545	28751	30776	58145	40960	46377	21861	25122	12562	36918	8826	27460	39665	59142	44176	30709	36677	25883	19223	11328	26312	26530	0	44933	21596
39679	22603	63767	81983	31603	89093	91820	11624	15967	50451	31595	42814	94947	43651	24756	77642	61298	68722	54886	18608	16595	67687	14566	65836	61767	60064	33084	34326	17278	45288	57980	75171	89147	49831	75516	28779	21512	27343	44937	26919	33392	44933	0	45551
44837	23202	62486	38946	62977	49975	49166	35487	35608	31842	29995	41657	49404	12138	21313	79082	23521	33402	41163	39282	30127	51307	55007	22493	25109	17206	13068	16592	45729	30192	12431	29697	44619	63213	42894	23765	35026	20380	10359	39486	42822	21596	45551	0

colnames(jakarta_district_centroid_expanded)

 [1] "province"                 "city"                    
 [3] "district"                 "centroid_lat"            
 [5] "centroid_lng"             "Cultural_Attractions"    
 [7] "Essentials"               "Facilities_Services"     
 [9] "Offices_Business"         "Others"                  
[11] "Restaurants_Food"         "Shops"                   
[13] "Recreation_Entertainment" "Tourism_Spots"           
[15] "geometry"                 "centroid"                

6.1.1 Pivot Distance Value by Origin and Destination District

distPair <- melt(dist_districts) %>%
  rename(dist = value)
head(distPair, 10)

               Var1   Var2     dist
1            cakung cakung     0.00
2     cempaka putih cakung 33262.49
3        cengkareng cakung 94446.12
4          cilandak cakung 81967.21
5         cilincing cakung 28291.84
6          cipayung cakung 64521.55
7           ciracas cakung 71798.66
8      danau sunter cakung 41185.87
9  danau sunter dll cakung 46710.36
10      duren sawit cakung 21978.34

There are intrazonal distances with 0 values (interzones)

Before proceeding, we update them with the lowest non-zero distance values before applying log transformation given that log 0 is an undefined value

distPair %>% 
  filter(dist > 0) %>% 
  summary()

            Var1                 Var2           dist       
 cakung       :  43   cakung       :  43   Min.   :  4074  
 cempaka putih:  43   cempaka putih:  43   1st Qu.: 28445  
 cengkareng   :  43   cengkareng   :  43   Median : 44576  
 cilandak     :  43   cilandak     :  43   Mean   : 46934  
 cilincing    :  43   cilincing    :  43   3rd Qu.: 62941  
 cipayung     :  43   cipayung     :  43   Max.   :125149  
 (Other)      :1634   (Other)      :1634                   

The lowest minimum distance is 4074 m, any distance less than this can represent the intrazonal distance. For consistency and ease of remberance, we stick to 2000m as the intra-zonal distance

distPair$dist <- ifelse (distPair$dist == 0, 
                         2000, distPair$dist) 
distPair <- distPair %>% 
  rename(origin=Var1, 
         destination =Var2) %>% 
  mutate(across(c(origin, destination), as.factor))

summary(distPair)

           origin            destination        dist       
 cakung       :  44   cakung       :  44   Min.   :  2000  
 cempaka putih:  44   cempaka putih:  44   1st Qu.: 27330  
 cengkareng   :  44   cengkareng   :  44   Median : 43652  
 cilandak     :  44   cilandak     :  44   Mean   : 45912  
 cilincing    :  44   cilincing    :  44   3rd Qu.: 62485  
 cipayung     :  44   cipayung     :  44   Max.   :125149  
 (Other)      :1672   (Other)      :1672                   

6.2 Applying Log Transformation to Variables of Interest

We apply log transformation for the variables of interest (categories of POI for push and pull factors), to use Poisson Regresion method.

1. For the subset of POIs in Origin and Destination Districts, we rename accordingly

poi_aggregated_origin <- poi_aggregated %>%
  rename(origin = district)

# Rename district to 'destination' for merging with destination data
poi_aggregated_destination <- poi_aggregated %>%
  rename(destination = district)

2. We merge the distPair with POI data for both origin and destination subsets
3. We Pivot the POI data wide to separate the categories into individual columns & counts

dist_with_POI <- distPair %>%
  left_join(poi_aggregated_origin, by = "origin") %>%
  left_join(poi_aggregated_destination, by = "destination", suffix = c("_origin", "_destination"))

poi_origin_wide <- dist_with_POI %>%
  pivot_wider(
    names_from = category_origin,
    values_from = count_origin,
    names_prefix = "origin_",
    values_fill = list(count_origin = 0)
  )

poi_orig_dest_wide <- poi_origin_wide %>%
  pivot_wider(
    names_from = category_destination,
    values_from = count_destination,
    names_prefix = "destination_",
    values_fill = list(count_destination = 0)
  )

4. We apply Log Transformation to the category columns

columns_to_log_transform <- colnames(poi_orig_dest_wide)[4:19]


poi_orig_dest_wide[columns_to_log_transform] <- lapply(
  poi_orig_dest_wide[columns_to_log_transform],
  function(x) log(as.numeric(as.character(x)) + 1)
)

head(poi_orig_dest_wide[columns_to_log_transform])

# A tibble: 6 × 16
  origin_Cultural_Attractions origin_Essentials origin_Facilities_Services
                        <dbl>             <dbl>                      <dbl>
1                       1.10               2.56                       3.14
2                       0.693              2.20                       2.56
3                       2.30               3.95                       4.23
4                       1.61               3.26                       4.04
5                       1.95               3.43                       3.71
6                       1.10               1.61                       3.00
# ℹ 13 more variables: origin_Offices_Business <dbl>, origin_Others <dbl>,
#   origin_Restaurants_Food <dbl>, origin_Shops <dbl>,
#   origin_Recreation_Entertainment <dbl>,
#   destination_Cultural_Attractions <dbl>, destination_Essentials <dbl>,
#   destination_Facilities_Services <dbl>, destination_Offices_Business <dbl>,
#   destination_Others <dbl>, destination_Restaurants_Food <dbl>,
#   destination_Shops <dbl>, destination_Recreation_Entertainment <dbl>

6.3 Origin-Constrained Spatial Interaction Model

The origin-constrained model focuses on constraints at the origin locations, meaning the sum of interactions originating from each origin equals a known total, such as the total number of trips from each origin district. Here, only pull (attractive) factors of the destination districts will be used.

Before doing so, we merge od_data and the newly created poi_orig_dest_wide

od_data <- od_data %>%
  rename(
    origin = origin_district,
    destination = destination_district
  )

tripsData <- left_join(od_data, poi_orig_dest_wide, by = c("origin", "destination"))

head(tripsData)

# A tibble: 6 × 27
  origin           destination origin_time_cluster origin_weather_descr…¹ trj_id
  <chr>            <chr>       <chr>               <chr>                  <chr> 
1 kelapa gading    pasar rebo  morning lull        not_rain               10003 
2 setia budi       pasar ming… morning peak        rain                   10043 
3 makasar          duren sawit morning peak        rain                   10061 
4 setia budi       mampang pr… morning peak        not_rain               10072 
5 kebayoran lama   pasar ming… morning lull        not_rain               10086 
6 grogol petambur… tanah abang midnight lull       not_rain               10089 
# ℹ abbreviated name: ¹​origin_weather_description_category
# ℹ 22 more variables: driving_mode <chr>, centroid_lat_origin <dbl>,
#   centroid_lng_origin <dbl>, centroid_lat_destination <dbl>,
#   centroid_lng_destination <dbl>, dist <dbl>,
#   origin_Cultural_Attractions <dbl>, origin_Essentials <dbl>,
#   origin_Facilities_Services <dbl>, origin_Offices_Business <dbl>,
#   origin_Others <dbl>, origin_Restaurants_Food <dbl>, origin_Shops <dbl>, …

tripsData_counts <- tripsData %>%
  group_by(origin, destination) %>%
  summarise(trips_count = n_distinct(trj_id), .groups = 'drop')  # Count unique trips

tripsData <- left_join(tripsData, tripsData_counts, by = c("origin", "destination"))

*For better viewing of variables in the 3 SIMs, we rename the origin and destination variables to read as such : “origin_district” or “deestination_district”

tripsData <- tripsData %>%
  rename(
    origin_ = origin,
    destination_ = destination
  )

origSIM <- glm(
  trips_count ~ origin_ + destination_Cultural_Attractions + destination_Essentials + 
                destination_Facilities_Services + destination_Offices_Business + 
                destination_Others + destination_Restaurants_Food + 
                destination_Shops + destination_Recreation_Entertainment + 
                dist - 1,  # No intercept
  family = poisson(link = "log"),
  data = tripsData,
  na.action = na.exclude
)


summary(origSIM)

Call:
glm(formula = trips_count ~ origin_ + destination_Cultural_Attractions + 
    destination_Essentials + destination_Facilities_Services + 
    destination_Offices_Business + destination_Others + destination_Restaurants_Food + 
    destination_Shops + destination_Recreation_Entertainment + 
    dist - 1, family = poisson(link = "log"), data = tripsData, 
    na.action = na.exclude)

Coefficients:
                                       Estimate Std. Error z value Pr(>|z|)    
origin_cakung                         1.757e+00  9.602e-02  18.299  < 2e-16 ***
origin_cempaka putih                  1.151e+00  8.098e-02  14.215  < 2e-16 ***
origin_cengkareng                     2.000e+00  6.744e-02  29.656  < 2e-16 ***
origin_cilandak                       1.824e+00  7.585e-02  24.048  < 2e-16 ***
origin_cilincing                      1.882e+00  1.104e-01  17.047  < 2e-16 ***
origin_cipayung                       2.171e+00  8.077e-02  26.883  < 2e-16 ***
origin_ciracas                        2.030e+00  7.427e-02  27.330  < 2e-16 ***
origin_danau sunter dll               8.483e-01  3.812e-01   2.225  0.02608 *  
origin_duren sawit                    2.455e+00  6.651e-02  36.915  < 2e-16 ***
origin_gambir                         1.463e+00  6.598e-02  22.168  < 2e-16 ***
origin_grogol petamburan              2.258e+00  5.847e-02  38.611  < 2e-16 ***
origin_jagakarsa                      2.312e+00  7.264e-02  31.827  < 2e-16 ***
origin_jatinegara                     1.918e+00  6.603e-02  29.041  < 2e-16 ***
origin_johar baru                     7.807e-01  1.168e-01   6.685 2.31e-11 ***
origin_kali deres                     1.832e+00  9.023e-02  20.305  < 2e-16 ***
origin_kebayoran baru                 2.257e+00  5.870e-02  38.456  < 2e-16 ***
origin_kebayoran lama                 2.055e+00  6.228e-02  32.996  < 2e-16 ***
origin_kebon jeruk                    2.072e+00  6.038e-02  34.309  < 2e-16 ***
origin_kelapa gading                  2.236e+00  6.263e-02  35.696  < 2e-16 ***
origin_kemayoran                      1.319e+00  7.927e-02  16.640  < 2e-16 ***
origin_kembangan                      2.002e+00  6.569e-02  30.479  < 2e-16 ***
origin_koja                           1.533e+00  9.134e-02  16.778  < 2e-16 ***
origin_kramat jati                    1.777e+00  7.280e-02  24.411  < 2e-16 ***
origin_makasar                        1.471e+00  1.023e-01  14.372  < 2e-16 ***
origin_mampang prapatan               1.117e+00  8.251e-02  13.535  < 2e-16 ***
origin_matraman                       9.069e-01  1.015e-01   8.931  < 2e-16 ***
origin_menteng                        1.415e+00  6.699e-02  21.116  < 2e-16 ***
origin_pademangan                     1.701e+00  7.487e-02  22.720  < 2e-16 ***
origin_palmerah                       1.292e+00  7.098e-02  18.204  < 2e-16 ***
origin_pancoran                       1.490e+00  7.305e-02  20.402  < 2e-16 ***
origin_pasar minggu                   2.458e+00  6.085e-02  40.394  < 2e-16 ***
origin_pasar rebo                     1.225e+00  1.114e-01  10.996  < 2e-16 ***
origin_penjaringan                    2.339e+00  6.396e-02  36.562  < 2e-16 ***
origin_pesanggrahan                   1.184e+00  9.893e-02  11.965  < 2e-16 ***
origin_pulo gadung                    1.696e+00  7.214e-02  23.506  < 2e-16 ***
origin_sawah besar                    1.216e+00  7.618e-02  15.961  < 2e-16 ***
origin_senen                          1.330e+00  7.272e-02  18.288  < 2e-16 ***
origin_setia budi                     2.202e+00  5.860e-02  37.582  < 2e-16 ***
origin_taman sari                     1.508e+00  7.130e-02  21.148  < 2e-16 ***
origin_tambora                        1.386e+00  7.208e-02  19.223  < 2e-16 ***
origin_tanah abang                    2.185e+00  5.996e-02  36.440  < 2e-16 ***
origin_tanjung priok                  2.142e+00  6.210e-02  34.482  < 2e-16 ***
origin_tebet                          2.022e+00  6.135e-02  32.952  < 2e-16 ***
destination_Cultural_Attractions      3.063e-02  1.501e-02   2.041  0.04120 *  
destination_Essentials                2.289e-01  1.813e-02  12.625  < 2e-16 ***
destination_Facilities_Services       6.091e-03  2.009e-02   0.303  0.76171    
destination_Offices_Business         -1.335e-02  7.795e-03  -1.713  0.08669 .  
destination_Others                    3.566e-02  1.249e-02   2.854  0.00431 ** 
destination_Restaurants_Food          9.069e-02  1.280e-02   7.084 1.40e-12 ***
destination_Shops                     9.246e-02  2.175e-02   4.252 2.12e-05 ***
destination_Recreation_Entertainment  9.041e-02  1.106e-02   8.173 3.02e-16 ***
dist                                 -4.832e-05  6.542e-07 -73.851  < 2e-16 ***
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

(Dispersion parameter for poisson family taken to be 1)

    Null deviance: 107406.8  on 3759  degrees of freedom
Residual deviance:   5523.5  on 3707  degrees of freedom
AIC: 19515

Number of Fisher Scoring iterations: 5

6.4 Destination-Constrained Spatial Interaction Model

In this model, we focus on the characteristics of origins that make them less attractive or less likely to generate demand for trips to destinations. Unlike the origin-constrained model, which emphasizes limitations or capacities at the origin, the destination-constrained model highlights the demand (attractiveness) of destinations by analyzing factors that make certain origins less likely to send trips.

destSIM <- glm(
  trips_count ~ destination_ + origin_Cultural_Attractions + origin_Essentials + 
                origin_Facilities_Services + origin_Offices_Business + 
                origin_Others + origin_Restaurants_Food + 
                origin_Shops + origin_Recreation_Entertainment + 
                dist - 1,  # No intercept
  family = poisson(link = "log"),
  data = tripsData,
  na.action = na.exclude
)

summary(destSIM)

Call:
glm(formula = trips_count ~ destination_ + origin_Cultural_Attractions + 
    origin_Essentials + origin_Facilities_Services + origin_Offices_Business + 
    origin_Others + origin_Restaurants_Food + origin_Shops + 
    origin_Recreation_Entertainment + dist - 1, family = poisson(link = "log"), 
    data = tripsData, na.action = na.exclude)

Coefficients:
                                  Estimate Std. Error z value Pr(>|z|)    
destination_cakung               2.360e+00  6.860e-02  34.397  < 2e-16 ***
destination_cempaka putih        1.357e+00  8.843e-02  15.350  < 2e-16 ***
destination_cengkareng           2.207e+00  6.253e-02  35.295  < 2e-16 ***
destination_cilandak             2.290e+00  6.641e-02  34.489  < 2e-16 ***
destination_cilincing            1.943e+00  9.876e-02  19.670  < 2e-16 ***
destination_cipayung             2.039e+00  9.427e-02  21.632  < 2e-16 ***
destination_ciracas              2.001e+00  8.865e-02  22.567  < 2e-16 ***
destination_danau sunter         4.959e-01  5.023e-01   0.987 0.323475    
destination_danau sunter dll    -7.051e-01  7.092e-01  -0.994 0.320111    
destination_duren sawit          2.339e+00  7.313e-02  31.979  < 2e-16 ***
destination_gambir               1.786e+00  5.935e-02  30.099  < 2e-16 ***
destination_grogol petamburan    2.003e+00  5.969e-02  33.561  < 2e-16 ***
destination_jagakarsa            2.755e+00  6.507e-02  42.342  < 2e-16 ***
destination_jatinegara           2.225e+00  6.151e-02  36.175  < 2e-16 ***
destination_johar baru           6.419e-01  1.802e-01   3.562 0.000368 ***
destination_kali deres           1.552e+00  1.013e-01  15.331  < 2e-16 ***
destination_kebayoran baru       2.421e+00  5.782e-02  41.868  < 2e-16 ***
destination_kebayoran lama       2.257e+00  6.043e-02  37.349  < 2e-16 ***
destination_kebon jeruk          1.871e+00  6.318e-02  29.620  < 2e-16 ***
destination_kelapa gading        1.840e+00  7.335e-02  25.080  < 2e-16 ***
destination_kemayoran            1.374e+00  7.707e-02  17.832  < 2e-16 ***
destination_kembangan            1.969e+00  6.540e-02  30.108  < 2e-16 ***
destination_koja                 1.853e+00  7.710e-02  24.032  < 2e-16 ***
destination_kramat jati          2.166e+00  6.987e-02  31.002  < 2e-16 ***
destination_makasar              1.732e+00  9.221e-02  18.778  < 2e-16 ***
destination_mampang prapatan     1.776e+00  6.621e-02  26.819  < 2e-16 ***
destination_matraman             1.232e+00  9.030e-02  13.641  < 2e-16 ***
destination_menteng              1.733e+00  6.134e-02  28.247  < 2e-16 ***
destination_pademangan           1.799e+00  6.963e-02  25.836  < 2e-16 ***
destination_palmerah             1.571e+00  6.456e-02  24.332  < 2e-16 ***
destination_pancoran             1.757e+00  6.936e-02  25.334  < 2e-16 ***
destination_pasar minggu         2.258e+00  6.494e-02  34.767  < 2e-16 ***
destination_pasar rebo           1.745e+00  8.416e-02  20.736  < 2e-16 ***
destination_penjaringan          2.482e+00  6.163e-02  40.267  < 2e-16 ***
destination_pesanggrahan         1.680e+00  7.868e-02  21.357  < 2e-16 ***
destination_pulo gadung          1.816e+00  6.962e-02  26.087  < 2e-16 ***
destination_sawah besar          1.329e+00  8.095e-02  16.419  < 2e-16 ***
destination_senen                1.324e+00  7.136e-02  18.549  < 2e-16 ***
destination_setia budi           2.253e+00  5.661e-02  39.790  < 2e-16 ***
destination_taman sari           9.732e-01  8.797e-02  11.063  < 2e-16 ***
destination_tambora              1.150e+00  8.140e-02  14.127  < 2e-16 ***
destination_tanah abang          2.218e+00  5.859e-02  37.859  < 2e-16 ***
destination_tanjung priok        2.137e+00  6.045e-02  35.354  < 2e-16 ***
destination_tebet                2.033e+00  6.064e-02  33.530  < 2e-16 ***
origin_Cultural_Attractions      5.248e-02  1.516e-02   3.462 0.000535 ***
origin_Essentials                3.353e-01  1.862e-02  18.007  < 2e-16 ***
origin_Facilities_Services      -1.772e-01  1.889e-02  -9.381  < 2e-16 ***
origin_Offices_Business          8.820e-02  8.129e-03  10.850  < 2e-16 ***
origin_Others                    1.025e-01  1.253e-02   8.178 2.89e-16 ***
origin_Restaurants_Food          8.175e-02  1.318e-02   6.201 5.61e-10 ***
origin_Shops                    -1.214e-02  2.078e-02  -0.584 0.559137    
origin_Recreation_Entertainment  3.943e-02  1.144e-02   3.446 0.000570 ***
dist                            -4.997e-05  6.669e-07 -74.927  < 2e-16 ***
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

(Dispersion parameter for poisson family taken to be 1)

    Null deviance: 107406.8  on 3759  degrees of freedom
Residual deviance:   5592.6  on 3706  degrees of freedom
AIC: 19586

Number of Fisher Scoring iterations: 5

6.5 Doubly Constrained Spatial Interaction Model.

The Doubly Constrained Spatial Interaction Model considers both the influence of origins and destinations on trip flows, balancing “push” factors from origins and “pull” factors from destinations.

dbcSIM <- glm(formula = trips_count ~ origin_ + destination_ + dist, 
              family = poisson(link = "log"),
  data = tripsData,
  na.action = na.exclude)

summary(dbcSIM)

Call:
glm(formula = trips_count ~ origin_ + destination_ + dist, family = poisson(link = "log"), 
    data = tripsData, na.action = na.exclude)

Coefficients:
                                Estimate Std. Error z value Pr(>|z|)    
(Intercept)                    3.203e+00  9.601e-02  33.363  < 2e-16 ***
origin_cempaka putih          -4.075e-01  1.003e-01  -4.063 4.85e-05 ***
origin_cengkareng              3.106e-01  9.213e-02   3.371 0.000749 ***
origin_cilandak               -1.345e-01  9.759e-02  -1.378 0.168241    
origin_cilincing               1.114e-02  1.292e-01   0.086 0.931284    
origin_cipayung                3.830e-01  1.097e-01   3.493 0.000477 ***
origin_ciracas                 4.487e-02  1.020e-01   0.440 0.660069    
origin_danau sunter dll       -8.396e-01  3.870e-01  -2.170 0.030042 *  
origin_duren sawit             5.814e-01  8.933e-02   6.508 7.62e-11 ***
origin_gambir                 -4.428e-02  9.000e-02  -0.492 0.622735    
origin_grogol petamburan       6.695e-01  8.533e-02   7.846 4.28e-15 ***
origin_jagakarsa               4.331e-01  9.754e-02   4.440 9.00e-06 ***
origin_jatinegara              1.157e-01  8.872e-02   1.304 0.192394    
origin_johar baru             -7.352e-01  1.313e-01  -5.601 2.14e-08 ***
origin_kali deres             -4.729e-03  1.120e-01  -0.042 0.966326    
origin_kebayoran baru          4.409e-01  8.557e-02   5.152 2.58e-07 ***
origin_kebayoran lama          2.332e-01  8.804e-02   2.649 0.008083 ** 
origin_kebon jeruk             3.285e-01  8.797e-02   3.735 0.000188 ***
origin_kelapa gading           5.258e-01  8.776e-02   5.992 2.08e-09 ***
origin_kemayoran              -3.082e-01  9.898e-02  -3.114 0.001848 ** 
origin_kembangan               3.949e-01  9.045e-02   4.365 1.27e-05 ***
origin_koja                   -2.494e-01  1.122e-01  -2.224 0.026149 *  
origin_kramat jati            -3.451e-02  9.452e-02  -0.365 0.715013    
origin_makasar                -4.576e-01  1.200e-01  -3.813 0.000137 ***
origin_mampang prapatan       -7.178e-01  1.022e-01  -7.020 2.22e-12 ***
origin_matraman               -8.298e-01  1.167e-01  -7.111 1.15e-12 ***
origin_menteng                -2.166e-01  8.986e-02  -2.410 0.015950 *  
origin_pademangan              1.247e-01  9.598e-02   1.299 0.193935    
origin_palmerah               -3.711e-01  9.318e-02  -3.982 6.83e-05 ***
origin_pancoran               -3.307e-01  9.426e-02  -3.508 0.000451 ***
origin_pasar minggu            5.254e-01  8.826e-02   5.953 2.64e-09 ***
origin_pasar rebo             -7.715e-01  1.299e-01  -5.939 2.86e-09 ***
origin_penjaringan             8.009e-01  8.878e-02   9.021  < 2e-16 ***
origin_pesanggrahan           -7.365e-01  1.179e-01  -6.246 4.21e-10 ***
origin_pulo gadung            -3.143e-02  9.417e-02  -0.334 0.738523    
origin_sawah besar            -3.119e-01  9.701e-02  -3.215 0.001305 ** 
origin_senen                  -2.733e-01  9.313e-02  -2.935 0.003338 ** 
origin_setia budi              4.776e-01  8.446e-02   5.654 1.57e-08 ***
origin_taman sari             -7.840e-02  9.273e-02  -0.845 0.397838    
origin_tambora                -1.637e-01  9.455e-02  -1.731 0.083468 .  
origin_tanah abang             5.135e-01  8.464e-02   6.067 1.31e-09 ***
origin_tanjung priok           6.098e-01  8.885e-02   6.864 6.71e-12 ***
origin_tebet                   2.474e-01  8.552e-02   2.892 0.003823 ** 
destination_cempaka putih     -7.762e-01  8.643e-02  -8.981  < 2e-16 ***
destination_cengkareng         1.315e-01  6.191e-02   2.124 0.033645 *  
destination_cilandak           3.037e-02  6.413e-02   0.474 0.635828    
destination_cilincing         -3.308e-01  9.922e-02  -3.334 0.000857 ***
destination_cipayung          -2.144e-01  9.910e-02  -2.163 0.030535 *  
destination_ciracas           -5.476e-01  1.005e-01  -5.450 5.03e-08 ***
destination_danau sunter      -1.929e+00  5.029e-01  -3.836 0.000125 ***
destination_danau sunter dll  -2.753e+00  7.094e-01  -3.881 0.000104 ***
destination_duren sawit        1.093e-01  7.084e-02   1.543 0.122824    
destination_gambir            -1.544e-01  5.745e-02  -2.687 0.007208 ** 
destination_grogol petamburan  8.412e-02  5.837e-02   1.441 0.149552    
destination_jagakarsa          4.718e-01  6.575e-02   7.176 7.18e-13 ***
destination_jatinegara         2.465e-02  5.600e-02   0.440 0.659860    
destination_johar baru        -1.332e+00  1.785e-01  -7.461 8.58e-14 ***
destination_kali deres        -5.012e-01  1.013e-01  -4.948 7.49e-07 ***
destination_kebayoran baru     3.172e-01  5.590e-02   5.674 1.39e-08 ***
destination_kebayoran lama     2.051e-01  5.870e-02   3.495 0.000474 ***
destination_kebon jeruk       -2.282e-01  6.206e-02  -3.677 0.000236 ***
destination_kelapa gading     -3.097e-01  7.089e-02  -4.369 1.25e-05 ***
destination_kemayoran         -7.134e-01  7.362e-02  -9.690  < 2e-16 ***
destination_kembangan          1.620e-02  6.478e-02   0.250 0.802565    
destination_koja              -5.040e-01  7.637e-02  -6.600 4.12e-11 ***
destination_kramat jati       -1.293e-01  6.887e-02  -1.877 0.060535 .  
destination_makasar           -6.126e-01  9.286e-02  -6.597 4.21e-11 ***
destination_mampang prapatan  -3.856e-01  6.304e-02  -6.116 9.58e-10 ***
destination_matraman          -9.475e-01  8.734e-02 -10.849  < 2e-16 ***
destination_menteng           -2.853e-01  5.918e-02  -4.820 1.43e-06 ***
destination_pademangan        -2.906e-01  6.688e-02  -4.346 1.39e-05 ***
destination_palmerah          -4.769e-01  6.192e-02  -7.702 1.34e-14 ***
destination_pancoran          -4.289e-01  6.653e-02  -6.446 1.15e-10 ***
destination_pasar minggu       1.456e-01  6.521e-02   2.232 0.025601 *  
destination_pasar rebo        -6.262e-01  8.607e-02  -7.276 3.44e-13 ***
destination_penjaringan        5.179e-01  5.917e-02   8.752  < 2e-16 ***
destination_pesanggrahan      -5.033e-01  7.834e-02  -6.425 1.32e-10 ***
destination_pulo gadung       -4.349e-01  6.679e-02  -6.511 7.45e-11 ***
destination_sawah besar       -7.267e-01  7.841e-02  -9.267  < 2e-16 ***
destination_senen             -6.100e-01  6.843e-02  -8.915  < 2e-16 ***
destination_setia budi         2.413e-01  5.463e-02   4.417 9.99e-06 ***
destination_taman sari        -9.770e-01  8.563e-02 -11.410  < 2e-16 ***
destination_tambora           -8.445e-01  7.844e-02 -10.767  < 2e-16 ***
destination_tanah abang        2.175e-01  5.464e-02   3.982 6.84e-05 ***
destination_tanjung priok      7.956e-02  6.002e-02   1.326 0.184978    
destination_tebet             -4.156e-02  5.709e-02  -0.728 0.466646    
dist                          -5.258e-05  6.921e-07 -75.971  < 2e-16 ***
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

(Dispersion parameter for poisson family taken to be 1)

    Null deviance: 20530  on 3758  degrees of freedom
Residual deviance:  3537  on 3672  degrees of freedom
AIC: 17598

Number of Fisher Scoring iterations: 5

7.0 Push-Pull Factor Analysis

In our prototyping and hypothesis, there were discussions on whether the number of category counts affected the push and pull factors of that particular district. For example, if a district had was populated with restaurants_and_food POIs as a category, compared to the origin district, we can theorize that residents or tourists might be going there because of said Point of Interest. While we recognize correlation is not causation, we will thus perform push-pull factor analysis to see the distribution and likely reasons why Grab demand might be generated from origin to destination.

7.1 Bar Chart of Coefficients

Steps: 1. We combine coefficients from models into a data frame

origin_constrained_coef <- coef(origSIM)[c("destination_Cultural_Attractions", "destination_Essentials", 
                                           "destination_Facilities_Services", "destination_Offices_Business", 
                                           "destination_Others", "destination_Restaurants_Food", 
                                           "destination_Shops", "destination_Recreation_Entertainment")]

destination_constrained_coef <- coef(destSIM)[c("origin_Cultural_Attractions", "origin_Essentials", 
                                                "origin_Facilities_Services", "origin_Offices_Business", 
                                                "origin_Others", "origin_Restaurants_Food", 
                                                "origin_Shops", "origin_Recreation_Entertainment")]

doubly_constrained_coef <- coef(dbcSIM)[c("origin_Cultural_Attractions", "origin_Essentials", 
                                          "origin_Facilities_Services", "origin_Offices_Business", 
                                          "origin_Others", "origin_Restaurants_Food", 
                                          "origin_Shops", "origin_Recreation_Entertainment")]

coefficients_df <- data.frame(
  Category = c("Cultural_Attractions", "Essentials", "Facilities_Services", 
               "Offices_Business", "Others", "Restaurants_Food", 
               "Shops", "Recreation_Entertainment"),
  Origin_Constrained = origin_constrained_coef,
  Destination_Constrained = destination_constrained_coef,
  Doubly_Constrained = doubly_constrained_coef
)

2. We melt the data for ggplot and then plot the coefficients

coefficients_long <- coefficients_df %>%
  pivot_longer(cols = -Category, names_to = "Model", values_to = "Coefficient")

# Plot
ggplot(coefficients_long, aes(x = Category, y = Coefficient, fill = Model)) +
  geom_bar(stat = "identity", position = "dodge") +
  theme_minimal() +
  labs(title = "Push-Pull Factor Coefficients by Model",
       x = "POI Category",
       y = "Coefficient")

3. We filter data for each model

origin_constrained_data <- coefficients_long %>% filter(Model == "Origin_Constrained")
destination_constrained_data <- coefficients_long %>% filter(Model == "Destination_Constrained")

num_categories <- length(unique(coefficients_long$Category))
category_colors <- brewer.pal(min(num_categories, 9), "Pastel1")
if (num_categories > 9) {
  category_colors <- colorRampPalette(brewer.pal(9, "Pastel1"))(num_categories)
}

4. Create individual plots for the model. For ease of visualization we rotate the axis and have a centered 0 axis.

plot_origin <- ggplot(origin_constrained_data, aes(y = Category, x = Coefficient, fill = Category)) +
  geom_bar(stat = "identity") +
  theme_minimal() +
  labs(title = "Origin-Constrained Model",
       x = "Coefficient",
       y = "POI Category") +
  theme(axis.text.y = element_text(angle = 0, hjust = 1),
        legend.position = "none") +  # Remove legend to avoid repetition in the plots
  scale_x_continuous(position = "top", limits = c(-max(abs(origin_constrained_data$Coefficient)),
                                                  max(abs(origin_constrained_data$Coefficient)))) +
  scale_fill_manual(values = category_colors)

plot_destination <- ggplot(destination_constrained_data, aes(y = Category, x = Coefficient, fill = Category)) +
  geom_bar(stat = "identity") +
  theme_minimal() +
  labs(title = "Destination-Constrained Model",
       x = "Coefficient",
       y = "POI Category") +
  theme(axis.text.y = element_text(angle = 0, hjust = 1),
        legend.position = "none") +  # Remove legend for consistency
  scale_x_continuous(position = "top", limits = c(-max(abs(destination_constrained_data$Coefficient)),
                                                  max(abs(destination_constrained_data$Coefficient)))) +
  scale_fill_manual(values = category_colors)

# Arrange the individual plots in a single column
grid.arrange(plot_origin, plot_destination, ncol = 1)

Other than the comparative Bar Charts above which effectively plots the coefficient of the attractiveness and propulsiveness of the POI categories, we can also consider Chord Diagrams to display the connections between origin and destination POI categories 1. Summarize trips data to get total trips between each origin and destination 2. Clear Previous Plots

chord_data <- tripsData %>%
  group_by(origin_, destination_) %>%
  summarise(trips_count = sum(trips_count)) %>%
  ungroup()

`summarise()` has grouped output by 'origin_'. You can override using the
`.groups` argument.

category_colors <- colorRampPalette(c("lightblue", "lightpink", "lightgreen", "lightcoral", "lightyellow"))(length(unique(c(chord_data$origin_, chord_data$destination_))))


circos.clear()

3. Plot Chord Diagram

chordDiagram(
  chord_data,
  grid.col = category_colors,        
  transparency = 0.3,                
  annotationTrack = "grid",           
  annotationTrackHeight = mm_h(5),    
  preAllocateTracks = list(track.height = 0.05)  #
)


circos.trackPlotRegion(track.index = 1, panel.fun = function(x, y) {
  sector.index <- get.cell.meta.data("sector.index")
  circos.text(CELL_META$xcenter, CELL_META$ylim[1] + mm_y(5), sector.index, facing = "clockwise", niceFacing = TRUE, adj = c(0, 0.5), cex = 0.8)
}, bg.border = NA)

title("Push-Pull Factors by Category in Chord Diagram", cex.main = 1.5)

head(top_districts)

[1] "setia budi"        "grogol petamburan" "kebayoran baru"   
[4] "penjaringan"       "tanah abang"      

head(poi_aggregated_top)

# A tibble: 6 × 3
  district          category                 count
  <chr>             <chr>                    <int>
1 grogol petamburan Cultural_Attractions         8
2 grogol petamburan Essentials                  59
3 grogol petamburan Facilities_Services        130
4 grogol petamburan Offices_Business            95
5 grogol petamburan Others                       4
6 grogol petamburan Recreation_Entertainment     6

8.0 Proposed Visualizations for OD Desire Line Maps

8.1 Additional Visualization for Trip Count by Time of Day (Heatmap)

To visualize the distribution of trips by time of day, we can use the code chunk below:

ORIGIN

tripsData$origin_time_cluster <- factor(
  tripsData$origin_time_cluster,
  levels = c("midnight peak", "midnight lull", "morning peak", "morning lull", 
             "afternoon peak", "afternoon lull", "evening peak", "evening lull")
)


ggplot(tripsData, aes(x = origin_time_cluster, y = origin_, fill = trips_count)) +
  geom_tile(color = "white") +  
  scale_fill_gradient(low = "white", high = "black", name = "Trip Count") +  
  scale_x_discrete(position = "top") +  
  labs(
    title = "Heatmap of Trips Throughout the Time of Day",
    x = "Time Cluster",
    y = "Destination District"
  ) +
  theme_void() +  
  theme(
    axis.text.x = element_text(angle = 0, hjust = 0.5),  
    axis.text.y = element_text(margin = margin(r = 5)), 
    plot.title = element_text(hjust = 0.5)  
  )

DESTINATION

tripsData$origin_time_cluster <- factor(
  tripsData$origin_time_cluster,
  levels = c("midnight peak", "midnight lull", "morning peak", "morning lull", 
             "afternoon peak", "afternoon lull", "evening peak", "evening lull")
)


ggplot(tripsData, aes(x = origin_time_cluster, y = destination_, fill = trips_count)) +
  geom_tile(color = "white") +  
  scale_fill_gradient(low = "white", high = "black", name = "Trip Count") +  # Grayscale palette
  scale_x_discrete(position = "top") +  
  labs(
    title = "Heatmap of Trips Throughout the Time of Day",
    x = "Time Cluster",
    y = "Destination District"
  ) +
  theme_void() +  # Removes all default grid elements
  theme(
    axis.text.x = element_text(angle = 0, hjust = 0.5),  
    axis.text.y = element_text(margin = margin(r = 5)),  
    plot.title = element_text(hjust = 0.5)  
  )

8.2 Split Plot Visualization for Weather Conditions:

1. We create a central 0 line
2. Represent ‘Not_Rain’ with Grey and ‘Rain’ with Blue

trip_counts_district <- tripsData %>%
  group_by(origin_, origin_weather_description_category) %>%
  summarise(trip_count = n(), .groups = "drop")


trip_counts_district <- trip_counts_district %>%
  mutate(trip_count = ifelse(origin_weather_description_category == "rain", -trip_count, trip_count))


ggplot(trip_counts_district, aes(y = origin_, x = trip_count, fill = origin_weather_description_category)) +
  geom_bar(stat = "identity", width = 0.8) +
  labs(
    title = "Trip Counts by Rain Condition per District",
    y = "District",
    x = "Trip Count",
    fill = "Weather Condition"
  ) +
  theme_minimal() +
  theme(
    axis.text.y = element_text(size = 8, margin = margin(r = 10)),  
    panel.grid = element_blank()                                 
  ) +
  scale_fill_manual(values = c("darkgrey", "blue")) +                 
  scale_x_continuous(labels = abs, limits = c(-max(trip_counts_district$trip_count), max(trip_counts_district$trip_count))) + geom_vline(xintercept = 0, color = "black") 

8.3 Shiny Prototype Sample: Time Cluster (with Vehicle Type)

8.4 Shiny Prototype Sample: Weather Condition with Split Plot Visualization

8.5 Shiny Prototype Sample: Points of Interest + Push and Pull Factor Analysis

Note

(To be Combined with Selected Desire Line Map Conditions)
For Prototyping Purposes we depict a singular point to see the tooltip and distribution of categories

For the Chord Diagram, Upon Clicking on a Specific District in the Radius, the user can visualize the flows of trips, in this Case District as Origin

9.0 Conclusion

In this project, I set out to understand the demand for Grab’s ride-hailing services in Jakarta by digging into spatial patterns and exploring how factors like time of day, weather, and vehicle type impact trip flows across the city. By analyzing demand patterns around specific Points of Interest (like restaurants, entertainment spots, and essential services), I found that these areas attract more trips during peak times and when it rains—insights that could help Grab, urban planners, and policymakers better manage the challenges of Jakarta’s traffic.

Geospatial Analysis, especially Spatial Interaction Modelling can show how smart data use can make city life smoother. By using visualizations like heatmaps and chord diagrams, I tried to break down complex mobility patterns into something more accessible. The insights here could help shape practical solutions for Jakarta’s traffic congestion, from adjusting service availability during high-demand times to making it easier for people to reach essential services without exacerbating traffic congestion. While this analysis may not offer direct solutions, it does offer insights based on complex geospatial variables.